摘要： 汉字签名鉴别是通过对手写人的笔迹进行分析从而得出签名是否由某特定人书写的结论。不同人的签名特征是不相同的，可以从其签名信息中抽取出最有效的签名特征来区分真假签名。该文采用新的数学工具——Wavelet 变换方法，利用复信号小波变换的优良特性抽取了一些新的具有旋转、平移及尺度不变性的签名特征，经验证这种特征具有很好的稳定性与可靠性。

关键词: 小波变换, 多分辩率分析, Mallat分解, 特征抽取

Abstract: Automatic signature verification about Chinese character is to draw a conclusion if this Chinese character is written by a person specially. The signatures of different individuals will be different, from those difference people can extract some effective signing features to classify an unknown signature as either authentic or forged.The paper adopts a new tool——Wavelet transform to extract some rotating-invariant,translating-invariant and dilating-invariant features,which have good stability and reliability according to the experiments. Taking advantage of this algorithm, a high positive-verification rate is obtained.

中图分类号: 

• TP391.43